fraggle_laptop has quit [Read error: Connection reset by peer]
fraggle_laptop has joined #libreelec
Klojum has joined #libreelec
Klojum has quit [Client Quit]
<trn>
Hello - I'm getting some extremely bizarre behavior on LibreELEC on an Rpi3, it runs for about 24 hours and then locks up - happens even with no services or kodi running, just copying data over SSH/SFTP ... it takes 5-10 seconds to respond to echo back typing when this happens, before it dies. Tunning anything triggers the OOM killer but almost no processes are running.
<trn>
I managed to grab the dmesg when this happened. I had a little swap on a hard disk and tried to use swapoff and it was OOM killed.
<trn>
Doesn't make a lot of sense to me as only 9 processes are running. This is 9.0.1. Before I try a new Pi and SD card to rule out the hardware, any idea what might possibly do that? Seems crazy to me.
<trn>
This is reproducible - copy a large amount of data via SFTP, it locks up in ~18-30 hours.
<trn>
Milhouse ? maybe ^
<trn>
Anyway thanks for all your hard work on LibreELEC, it's great.
<Milhouse>
@trn: Maybe it's this issue https://github.com/raspberrypi/linux/issues/2829 - can you reproduce with one of my test builds with kernel 5.0.3? Although LibreELEC isn't using the same kernel config settings as Raspbian so it could be a completely different issue.
<trn>
Milhouse: Good morning! I'll give it a try later and let you know.
<trn>
Milhouse: It's been going fine for awhile now, but I'm still running the script I pasted above. It's not easy for me to visually parse slabinfo, but so far looking at slabinfo it appears saneZ
<trn>
*sane. It's hard to catch the problem because it happens only after 18-36 hrs, and the only way I know it's about to fall apart is to see the transfer speed start going down
<trn>
From ~12M to 2 to 1 ... I have about 5 minutes once it starts slowing while it's still usable to get some debugging info before it's fully hung, makes debugging it hit or miss.
<trn>
It doesn't actually ever die completely, sometimes when it seems hung you can wait ~30m and another 1K comes through, ends up taking >45m to echo a key back over SSH (or to even just login).
<trn>
I still think it's more related to fragmentation than caching - but my drop_caches loop would mitigate both.
<trn>
Also, on another Pi I'm running 4.19.30 without an issue but didn't notice the issue when I was running older versions, though.
<trn>
For LibreELEC use it isn't a problem to run with all the caches essentially disabled. But, that other issue you pointed at about the ext4_inode_cache is potentially relevant as well - I am using ext4 for the local disks.
<trn>
Thanks Milhouse.
baggar11 has joined #libreelec
RaphGro has joined #libreelec
RaphGro has quit [Client Quit]
ca_cabotage has quit [Ping timeout: 258 seconds]
ca_cabotage has joined #libreelec
hijacker has quit [Quit: Leaving]
tuxiano has joined #libreelec
TheTrash has quit [Remote host closed the connection]
TomTom has quit [Quit: Connection closed for inactivity]
fraggle_laptop has quit [Remote host closed the connection]
fraggle_laptop has joined #libreelec
TheTrash has joined #libreelec
BenG83 has quit [Ping timeout: 255 seconds]
|Jeroen| has joined #libreelec
tuxiano has quit [Quit: tuxiano]
tuxiano has joined #libreelec
gouchi has joined #libreelec
aphirst has left #libreelec ["Leaving"]
antlarr has quit [Quit: bye!]
antlarr has joined #libreelec
antlarr has quit [Quit: bye!]
antlarr has joined #libreelec
tuxiano has quit [Quit: tuxiano]
fysa has joined #libreelec
<fysa>
looking for testers for https://github.com/digitalsanity/libreelec-in-docker/ .. grabs/runs latest nightly libreelec front-end as a docker service. Update config to point to the build URL and run install.sh from any debian/ubuntu environment -- e.g. lets you run libreelec on an OMV box or while other host services are running and easy upgrades without reflashing
<trn>
fysa cool project, saw you mention it yesterday
<fysa>
also have recalbox working :)
<fysa>
would be possible to make a recalbox container installable via libreelec docker add-on
antlarr has quit [Quit: bye!]
<fysa>
would be nice to have a couple odroid testers :)
antlarr has joined #libreelec
tommy`` has quit [Ping timeout: 245 seconds]
antlarr has quit [Client Quit]
antlarr has joined #libreelec
antlarr has quit [Client Quit]
antlarr has joined #libreelec
antlarr has quit [Client Quit]
tommy`` has joined #libreelec
antlarr has joined #libreelec
ca_cabotage has quit [Ping timeout: 245 seconds]
<TheTrash>
I'm fiddling around with the Chrome addon on x86 to tweak the UI settings for TV use a bit, but I'm having a hard time figuring out where to start changing stuff like the UI font, window decoration settings, etc. -- would I set it through xorg, or through gtk, or pango, or something entirely else? And what would the config file paths be, running from the addon dir?
<TheTrash>
I don't entirely understand what it shares with the running xorg session and not. E.g. if it's the same session Kodi was running in, or a new one.
|Jeroen| has quit [Quit: dada]
jedineeper has quit [Quit: Connection closed for inactivity]
BenG83 has joined #libreelec
ghostcube has quit [Quit: Verlassend]
shootbird has quit [Quit: No Ping reply in 180 seconds.]
shootbird has joined #libreelec
fraggle_laptop has quit [Read error: Connection reset by peer]
fraggle_laptop has joined #libreelec
Fenster` has joined #libreelec
Fenster has quit [Ping timeout: 250 seconds]
Fenster` has quit [Ping timeout: 246 seconds]
benoliver999 has quit [Ping timeout: 268 seconds]
<MaDMaLKaV>
bad news is I only got one person to test my alpha build for a5x box. good news is the guy is testing a lot of stuff I couldn't
<MaDMaLKaV>
so, not bad
benoliver999 has joined #libreelec
shootbird has quit [Quit: No Ping reply in 180 seconds.]
shootbird has joined #libreelec
thesquib has quit [Ping timeout: 268 seconds]
thesquib has joined #libreelec
gouchi has quit [Remote host closed the connection]